A Yeast taf17 Mutant Requires the Swi6 Transcriptional Activator for Viability and Shows Defects in Cell Cycle-Regulated Transcription

Corresponding author: Brenda Andrews, Department of Molecular and Medical Genetics, University of Toronto, Rm. 4285, Medical Sciences Bldg., 1 King's College Circle, Toronto, Ontario M5S 1A8, Canada., brenda.andrews{at}utoronto.ca (E-mail)

Communicating editor: A. P. MITCHELL

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In Saccharomyces cerevisiae, the Swi6 protein is a component of two transcription factors, SBF and MBF, that promote expression of a large group of genes in the late G1 phase of the cell cycle. Although SBF is required for cell viability, SWI6 is not an essential gene. We performed a synthetic lethal screen to identify genes required for viability in the absence of SWI6 and identified 10 complementation groups of swi6-dependent lethal mutants, designated SLM1 through SLM10. We were most interested in mutants showing a cell cycle arrest phenotype; both slm7-1 swi6 and slm8-1 swi6 double mutants accumulated as large, unbudded cells with increased 1N DNA content and showed a temperature-sensitive growth arrest in the presence of Swi6. Analysis of the transcript levels of cell cycle-regulated genes in slm7-1 SWI6 mutant strains at the permissive temperature revealed defects in regulation of a subset of cyclin-encoding genes. Complementation and allelism tests showed that SLM7 is allelic with the TAF17 gene, which encodes a histone-like component of the general transcription factor TFIID and the SAGA histone acetyltransferase complex. Sequencing showed that the slm7-1 allele of TAF17 is predicted to encode a version of Taf17 that is truncated within a highly conserved region. The cell cycle and transcriptional defects caused by taf17^slm7-1 are consistent with the role of TAF_IIs as modulators of transcriptional activation and may reflect a role for TAF17 in regulating activation by SBF and MBF.

IN the budding yeast Saccharomyces cerevisiae, irreversible commitment to mitosis occurs late in the G1 phase of the cell cycle. Passage through the commitment phase, or START, depends on activation of the cyclin dependent kinase (Cdk) Cdc28 by association with the G1 cyclins Cln1, Cln2, and Cln3 (NASMYTH and DIRICK 1991 ; OGAS et al. 1991 ; DIRICK et al. 1992 ; LOWNDES et al. 1992 ; BREEDEN and MIKESELL 1994 ; CROSS et al. 1994 ; STUART and WITTENBERG 1994 ). In the absence of CLN1 and CLN2, the CDK Pho85 and its associated G1 cyclins Pcl1 and Pcl2 become essential for START (ESPINOZA et al. 1994 ; MEASDAY et al. 1994 ). Consistent with their role at START, CLN1, CLN2, PCL1, and PCL2 are periodically transcribed, with peak transcription in late G1 (WITTENBERG et al. 1990 ; FERNANDEZ-SARABIA et al. 1992 ; TYERS et al. 1993 ; MEASDAY et al. 1994 ). Recent global analysis of transcription patterns by DNA microarray hybridization reveals that CLN1, CLN2, PCL1, and PCL2 are coordinately expressed around the time of START with a large group of ~120 genes (CHO et al. 1998 ; SPELLMAN et al. 1998 ). Cell cycle-dependent transcription of this group of genes is governed by two heterodimeric transcription factors, called SBF and MBF. SBF contains the Swi4 and Swi6 proteins and usually activates transcription through a cis-acting element called the SCB (SWI4/6 cell cycle box; consensus CACGAAA) (NASMYTH 1985 ; BREEDEN and NASMYTH 1987A ; ANDREWS and HERSKOWITZ 1989A , ANDREWS and HERSKOWITZ 1989B ; NASMYTH and DIRICK 1991 ; OGAS et al. 1991 ; TABA et al. 1991 ; KOCH and NASMYTH 1994 ). The Swi4 protein specifically binds to SCB elements, whereas Swi6 does not bind DNA specifically (SIDOROVA and BREEDEN 1993 ) but is present in SBF by virtue of its interaction with the C terminus of Swi4 (ANDREWS and MOORE 1992 ; PRIMIG et al. 1992 ; SIDOROVA and BREEDEN 1993 ). In the absence of Swi4 or Swi6, G1 cyclin transcription is greatly reduced but still detectably periodic (BREEDEN and NASMYTH 1987A ; NASMYTH and DIRICK 1991 ; OGAS et al. 1991 ; DIRICK et al. 1992 ; LOWNDES et al. 1992 ; BREEDEN and MIKESELL 1994 ; CROSS et al. 1994 ; STUART and WITTENBERG 1994 ).

The Swi6 protein also interacts with a second DNA-binding protein, Mbp1, to form a distinct transcription factor, MBF (MCB-binding factor, also known as DSC1; DIRICK et al. 1992 ; KOCH et al. 1993 ; LOWNDES et al. 1991 , LOWNDES et al. 1992 ; VERMA et al. 1992 ). MBF/DSC1 activates START-specific expression of the SWI4 gene, the S-phase cyclin genes CLB5 and CLB6, as well as many genes needed for DNA synthesis such as RNR1 and TMP1 (DIRICK et al. 1992 ; LOWNDES et al. 1992 ; FOSTER et al. 1993 ; KOCH et al. 1993 ; SCHWOB and NASMYTH 1993 ). SBF- and MBF-dependent gene expression is strongly activated by the Cln3-Cdc28 kinase when cells reach a critical size at START (TYERS et al. 1993 ; DIRICK et al. 1995 ; STUART and WITTENBERG 1995 ). However, the precise mechanism of Cln3-Cdc28 action on SBF and MBF remains a mystery. After START, repression of the G1/S program is caused by the Cdc28 kinase in association with the B-type cyclins or Clbs. Cells that lack Clb1-4 are unable to repress SBF-dependent genes (AMON et al. 1993 ) and ~100 transcripts are repressed by overexpression of Clb2 (SPELLMAN et al. 1998 ). Repression of SBF-regulated genes may involve physical association of Swi4 with Clb2 during G2, although Mbp1 does not appear to interact directly with Clb2 (AMON et al. 1993 ; SIEGMUND and NASMYTH 1996 ).

Although neither SWI4 nor SWI6 is an essential gene, their importance is demonstrated by the fact that a cell deleted for both SWI4 and SWI6 arrests at START (BREEDEN and NASMYTH 1987A ). This "synthetic lethality" is due to reduced expression of the G1 cyclins since ectopic expression of CLN1, CLN2, or PCL1 rescues a swi4 swi6 strain (NASMYTH and DIRICK 1991 ; OGAS et al. 1991 ). Unlike SBF, MBF is a nonessential complex since a swi6 mbp1 cell is alive (KOCH et al. 1993 ). The viability of strains lacking MBF most likely reflects the fact that MCB-regulated genes are not downregulated in cells lacking MBF but instead are expressed constitutively throughout the cell cycle (KOCH et al. 1993 ). Although SWI6 is not an essential gene, at least one of the DNA-binding proteins associated with Swi6 is required for START, since a swi4 mbp1 cell dies prior to DNA synthesis (KOCH et al. 1993 ).

One implication of these observations is that there must be substantial cross talk between the SBF and MBF pathways in vivo. Indeed, MBF complex formation can be specifically competed by SCB oligomers in vitro, indicating that MBF may also bind to SCBs (LOWNDES et al. 1991 ; DIRICK et al. 1992 ). In addition, SBF binds to MCB elements in the CLN1 promoter in vitro and these elements depend on Swi4 and Swi6 for their in vivo function (PARTRIDGE et al. 1997 ). It is important to note, however, that in vivo footprinting experiments reveal that SCB sequences are not detectably protected from chemical modification in swi4 or swi6 strains (HARRINGTON and ANDREWS 1996 ; KOCH et al. 1996 ). In the absence of Swi6, Swi4 and Mbp1 may bind weakly to SCB and MCB elements and function to activate transcription at levels sufficient for cell viability.

A great deal of evidence suggests the existence of proteins that are functionally redundant or interact with Swi6. For example, transcription of the SWI4 gene itself is cell cycle-regulated with peak expression occurring slightly before that of HO (BREEDEN and MIKESELL 1991 ). Cell cycle-regulated expression of SWI4 is partially controlled by MCB sequences, but a substantial contribution to that regulation is MCB independent yet still dependent on SWI6 (FOSTER et al. 1993 ). This observation suggests the existence of another factor that may interact with Swi6 to regulate cell cycle-dependent transcription. The existence of regulators that are functionally redundant with Swi6 is indicated by the fact that swi6 deletion mutants are still alive despite the absence of both SBF and MBF activity. As discussed above, this can be explained by the residual, relatively nonperiodic expression of essential targets of SBF and MBF in swi6 strains. Thus, an alternative pathway for expression of SBF- and MBF-driven genes is important for cell viability in the absence of Swi6. We reasoned that if either the redundant pathway were nonfunctional or SWI4 were turned off, cells lacking SWI6 would die. To identify factors that maintain cell viabilility in the absence of SWI6, we have performed a genetic screen for mutations that create a requirement for Swi6.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains, media, and genetic methods:
Yeast strains used in these experiments are described in Table 1. Synthetic minimal media containing glucose or galactose (SD or SG) and rich media containing glucose or galactose (YEPD or YEPG) were prepared as described (SHERMAN and HICKS 1991 ). 5-Fluoroorotic acid (5-FOA) plates were SD or SG -Ura plates supplemented with 50 mg/l uracil and 1 mg/ml 5-FOA (Toronto Research Chemicals). Crosses, sporulation, and tetrad dissection were performed essentially as described (SHERMAN and HICKS 1991 ). Yeast transformations were performed using slight modifications of standard techniques (GIETZ et al. 1992 ).

NY7 (MATa slm6-1), NY11 (MATa slm7-1), NY15 (MATa slm8-1), and NY34 (MAT slm7-1) were recovered as meiotic segregants from diploids formed by mating each of NYS6 (swi6 slm6-1 [pNM1]), NYS7 (swi6 slm7-1 [pNM1]), and NYS8 (swi6 slm8-1 [pNM1]) to a wild-type strain (BY109). Segregants were tested for growth on medium lacking histidine and for death at 37° on YEPD. For construction of strain NY36 (TAF17::LEU2), plasmid pNM11, which contains a genomic fragment carrying the TAF17 gene in an integrating vector (see description below), was cleaved within the insert to direct homologous integration and transformed into a wild-type strain (JO34). As a result, the site of integration in the TAF17 gene was marked by the plasmid-borne LEU2 gene.

Plasmids:
To construct plasmid pNM1, a XhoI-BglII fragment containing the SWI6 gene (BREEDEN and NASMYTH 1987B ) was cloned into the XhoI-BamHI sites of plasmid pJO14. The pJO14 plasmid is a CEN/ARS-based shuttle vector derived from pRS316 (SIKORSKI and HIETER 1989 ) with a URA3 selectable marker and the GAL1-10 promoter inserted in the polylinker sequences. To construct a LEU2-SWI4 plasmid, a XhoI-SalI fragment containing the SWI4 gene and flanking sequences was isolated from clone B3.2 (ANDREWS and HERSKOWITZ 1989B ) and cloned into the SalI site of pUC18Bgl2 (a derivative of pUC18 with BglII linkers inserted at the EcoRI and HindIII sites) to create plasmid pBA313. The SWI4 gene was then isolated on a BglII fragment from pBA313 and cloned into BamHI-digested pRS315 to create plasmid pBA417. The pNM10 plasmid was constructed by inserting a genomic BamHI fragment containing TAF17 into the BamHI site of pRS315. The pNM11 plasmid was constructed by inserting the same genomic BamHI fragment containing TAF17 into pRS305 integrating vector (SIKORSKI and HIETER 1989 ).

Mutagenesis:
Strain NY3 was mutagenized by exposure to ethyl methanesulfonate (EMS; Sigma Chemical Co., St. Louis). Twelve cultures of NY3 from individual colonies were grown overnight in 2 ml of YEPG. The cultures were washed, resuspended in 1.5 ml of 0.1 M sodium phosphate buffer pH 7.0, and sonicated for 10 sec to separate clumpy cells. An aliquot (200 µl) was removed to serve as an untreated control and 20 µl of EMS was added to the remaining culture. The cells were incubated with aeration at 30° for 70 min and washed in 1.5 ml of 5% sodium thiosulfate to inactivate the EMS. The cells were then washed twice in 1.5 ml of water, resuspended in 1 ml of YEPG, and plated onto YEPG plates (100 µl/plate). This treatment resulted in ~50% cell survival.

Assays of mutant phenotypes:
The colonies derived from survivors of EMS treatment were replica plated onto two 5-FOA-containing plates and two YEPD plates. One set of plates was incubated at 25° and the other at 37° for 3 days. Colonies from each 5-FOA-containing or YEPD plate were then replica plated to YEPD and incubated for 2 days at the same temperature as the previous plates. Colonies that did not grow after treatment with 5-FOA were dead because the cells could not survive without Swi6. Cells that failed to grow on YEPD died because the levels of Swi6 produced were not enough to keep the swi6 slm mutant alive or because the cells could not metabolize glucose. Colonies that did not grow after the second incubation were streak purified from the original YEPG plates and patched onto YEPG at 25°. The resulting patches were tested for growth on 5-FOA plates containing galactose as the carbon source. Mutants that grew on these plates were judged to be defective in glucose metabolism and were not studied further.

Mutants that died on YEPD or 5-FOA-containing medium at 37° were assayed for growth on YEPG at 37°. Mutants that died under these conditions were judged to be temperature sensitive in the presence of high levels of Swi6 and were not studied further. Mutants that died at 25° were not assayed in this manner, since a mutation that is synthetically lethal with swi6 at 25° could result in temperature sensitivity, even in the presence of Swi6. An example of this type of mutation is a null mutation in SWI4, which is synthetically lethal with swi6 at 25°, but which causes mutant cells to die at 37°, even in the presence of wild-type SWI6 (BREEDEN and NASMYTH 1987A ; OGAS et al. 1991 ). Growth on YEPG at 37° would also not eliminate temperature-sensitive mutations that are suppressed by high levels of Swi6.

Cells that did not contain the pNM1 plasmid, or contained mutations in URA3, may be dying for reasons other than Swi6 levels. Mutant isolates that failed to grow on medium lacking uracil were not studied further. Mutants were also tested for rescue of FOA sensitivity by transformation with a SWI6 plasmid carrying the LEU2 nutritional marker; only those isolates rescued by transformation with the SWI6-LEU2 plasmid were studied.

Complementation and allelism tests:
The original MAT swi6 slm isolates were mated to BY168, an isogenic MATa swi6 strain (Table 1). The mutation was judged to be dominant if the diploid was dead under nonpermissive conditions. MATa swi6 slm mutant segregants were isolated from these diploids by analyzing meiotic progeny. The MATa swi6 slm isolates were mated to MAT swi6 strains containing a different slm mutation. The resulting diploids were tested under the most restrictive of the minimal conditions for death of the two mutants (25° on YEPD being least restrictive and 37° on 5-FOA being most restrictive) to assay for complementation of the slm mutant phenotype. If the diploid strain was unable to survive under the restrictive conditions, the slm mutants were judged to be in the same complementation group.

Representatives of each complementation group were backcrossed three times to BY165 (wild type) to obtain swi6 slm strains less likely to carry confounding secondary mutations. The resulting swi6 slm strains transformed with pNM1 are designated NYS1 through NYS10 and contain mutations in slm1 through slm10, respectively (see Table 1). The number of mutant strains in each complementation group was the following: slm1, 6 isolates; slm2, 6 isolates; slm3, 17 isolates; slm4, 2 isolates; slm5, 7 isolates; slm6, 18 isolates; slm7, 27 isolates; slm8, 15 isolates; slm9, 26 isolates; slm10, 4 isolates.

To test allelism of slm mutations with SWI4, NYS4 and NYS6 were mated to strain BY167 (swi4). The resulting diploid was sporulated and the segregation of lethality in meiotic progeny was assessed. The allelism of TAF17 with slm7 was tested by mating NY34 (MAT slm7-1) to NY36 (MATa TAF17::LEU2) and monitoring the segregation of leucine auxotrophy and temperature sensitivity (see RESULTS).

Cell cycle analysis:
For time course studies shown in Fig 3A, logarithmically growing cells of NYS6, NYS7, NYS8, or BY147 growing in SG medium at 25° were collected, resuspended to an OD₆₀₀ of 0.2 in SD medium, and grown at 37°. In Fig 3B, log phase cultures of NY7, NY11, NY15, BY108, and JO34 in SD medium at 30° were diluted to an OD₆₀₀ of 0.2 and grown in SD medium at 37°. Samples were removed from the cultures every 2 hr, fixed, and prepared for FACS analysis (MEASDAY et al. 1994 ). Where indicated, fixed cells were treated with 10 µg/ml DAPI (4',6-diamidino-2-phenylindole; Sigma Chemical Co.) and washed one to four times with water. The fixed cells were photographed at 400x magnification with a Nikon Microphot-FXA and Kodak TX-400 film or with a CCD camera mounted on a Leica DM-LB microscope. Images from the CCD camera were analyzed using a Northern Exposure Imaging System (Empix Imaging, Inc., Mississauga, Ontario). DAPI images were obtained using a 355–375-nm excitation filter and a 400-nm long pass barrier filter.

View larger version (124K): [in this window] [in a new window] [Download PPT slide]   Figure 1. slm mutants require SWI6 for viability. The indicated cultures were incubated at 25°, and serial dilutions of log phase cultures were spotted onto YEPG-, YEPD-, and 5-FOA-containing media. The plates containing strains that show synthetic lethality at 25° were incubated at 25°, whereas those containing strains that show synthetic lethality at 37° were incubated at 37°. NYS"X" indicates a slmX swi6 [pNM1] strain, BY147 is swi4 swi6 [pGAL::SWI4], BY109 is wild type, and NY3 is the parental swi6 strain.

View larger version (83K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Rescue of NYS4 and NYS6 by SWI4 and SWI6. NYS4 and NYS6 strains were transformed with pBA417 (SWI4) and pBA354 (SWI6). The resultant transformants were grown to OD600 = 0.1 and serial dilutions were spotted onto YEPD at 25° and 37° and grown for 6 days (25°) or 2 days (37°).

View larger version (189K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Analysis of arrest phenotypes of slm mutants. (A) Strains BY147 (swi4 swi6 pGAL::SWI6), NYS6 (swi6 slm6-1 pGAL::SWI6), NYS7 (swi6 slm7-1 pGAL::SWI6), and NYS8 (swi6 slm8-1 pGAL::SWI6) were grown to log phase in galactose-containing medium and then shifted to glucose-containing medium at 37° (nonpermissive conditions) for 8 hr. Samples were harvested every 2 hr, fixed and stained with propidium iodide for FACS analysis (top of A) or with DAPI for fluorescence microscopy. Photomicrographs of the DAPI-stained cells and Nomarski images of the cells are shown [DAPI and DIC (differential interference contrast) indicated to right of panel]. Time of incubation at 37° in glucose is shown to the right of the FACS profiles (0, 2, 4, 6, or 8 hr) and to the left of the photomicrographs (0, 4, and 8 hr). (B) FACS profile of NY7 (slm6-1 SWI6), NY11 (slm7-1 SWI6), NY15 (slm8-1 SWI6), BY108 (swi4 SWI6), and JO34 (wild type) grown to log phase at 30° in glucose and shifted to 37°. Cells were treated as described above for FACS analysis and photographed with Nomarski optics.

Northern analysis:
Logarithmically growing cells of JO34 (wild type) and NY11 (slm7-1) were treated for 165 min with 3 µM -factor, washed in ice-cold water, and resuspended in YEPD at 25°. Samples were collected at 10-min intervals over 200 min for JO34 and at 15-min intervals over 300 min for NY11, since the doubling time of NY11 is ~1.5x that of wild type. Total RNA was extracted from these samples as previously described (MEASDAY et al. 1997 ). An aliquot of 20 µg of RNA from each sample was run on a MOPS/formaldehyde gel and transferred to a nylon membrane. The RNA was immobilized on the membrane by UV crosslinking at 120 mJ/cm² with a Fisher Scientific UV crosslinker. Immobilized RNA was probed as described (MEASDAY et al. 1997 ) for PCL1, PCL2, SWI5, CLN2, ACT1, and PCL9 transcripts. NY11 and wild-type RNA immobilized on a fresh membrane was probed for CLB5, PCL1, and ACT1. All the membranes were treated identically for the entire procedure. The resulting membranes were exposed to a phosphorimager screen for 16–24 hr and the screen was scanned using a Molecular Dynamics (Sunnyvale, CA) phosphorimager. The transcripts were quantitated with Image-QuaNT (v4.0) software. Individual bands were quantitated by drawing a box around the band and using an identical box in the same lane as background. All bands were normalized to the ACT1 transcript and the resulting data were plotted using Microsoft Excel.

Measuring total mRNA in mutant cells:
Wild type (BY109), slm7-1 (NY11), taf145^ts (YSW93), and rbp1-1 (Y260) cells were grown to OD₆₀₀ = 0.2 at 30° and then shifted to 37°. Aliquots were taken at 0, 1, 2, and 4 hr after transfer to 37°. RNA was isolated from these pellets (see above), and 10 µg of RNA from each sample was spotted onto a nylon membrane (Hybond N⁺, Amersham, Buckinghamshire, UK) using a hybridot manifold (BRL). The blot was washed and then probed with ³²P-labeled oligo(dT)₂₀ as described in WALKER et al. 1997 .

Cloning of SLM7:
A yeast genomic library (ENGEBRECHT et al. 1990 ) was transformed into strain NY11 (slm7-1) and the resulting transformants were replica plated to SD medium and grown at 37° for 5 days. Plasmids were isolated from colonies that survived at 37° and rescue of the temperature-sensitive phenotype of NY11 was confirmed by retransformation of NY11 (slm7-1) with the appropriate plasmid.

The genomic insert of one of the NY11-complementing clones was sequenced and a fragment containing TAF17 was subcloned. Plasmid pNM10 (pRS315-TAF17) was then transformed into NY11 (slm7-1) and NYS7 (slm7-1 swi6). These transformants were tested for growth on YPD at 37° and on 5-FOA at 25°, respectively. Other rescuing plasmids were tested by PCR to determine whether they contained TAF17. Primers used to amplify TAF17 were 5' primer (T17-1), GAGGATCCTTATGAACGGCGGAGG and 3' primer (T17-2), GCGGATCCTCACATAGACTTTGGG. To confirm rescue of the slm7 mutant phenotype, NY11 (slm7-1) was transformed with pZM255 (MOQTADERI et al. 1996A ) and pNM10 (pRS315-TAF17). pZM255 is a plasmid containing the TAF17 gene under control of the GAL promoter. The resulting transformants were spotted on glucose- and galactose-containing media and grown at 30° and 37° for 2 days (glucose) or 6 days (galactose).

Sequencing of slm7-1:
The TAF17 locus in NY11 was amplified by PCR using primers T17-1 and T17-2 (see above). Both strands of three different PCR fragments resulting from this amplification were sequenced using primers T17-1 and T17-2.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Screen for swi6-dependent lethal mutants:
The Swi6 protein plays an important regulatory role in two transcription factors, SBF and MBF, that are required for the induction of gene expression at START in S. cerevisiae. Despite the fact that many of the genes controlled by SBF and MBF are essential for cell cycle progression, a swi6 deletion strain is viable. As such, it seemed likely that some gene(s) or pathway(s) exists in yeast cells that partially compensates for the lack of SWI6 function.

To identify genes involved in such a parallel pathway, and to identify genes that may be functioning in the same pathway as SBF, we devised a synthetic lethal screen for mutants that require SWI6 for viability. This strategy has been useful for identification of genes involved in regulating a common process (BENDER and PRINGLE 1991 ; COSTIGAN et al. 1992 ; GUARENTE 1993 ; HOLTZMAN et al. 1993 ; IGUAL et al. 1996 ). Our strategy for identifying genes that support growth of a strain lacking Swi6 was to mutagenize a swi6 strain transformed with pNM1, a centromere-based plasmid that contains the URA3 nutritional marker and the SWI6 gene expressed from the inducible GAL1-10 promoter. When a swi6 [pNM1] strain (NY3, Table 1) was grown on glucose-containing medium, expression of SWI6 from the GAL promoter was repressed to the extent that Swi6 was undetectable by Western blotting using anti-Swi6 antiserum (data not shown). However, a more sensitive biological assay revealed some residual SWI6 expression from the glucose-repressed GAL1-10 promoter on the pNM1 plasmid. Whereas a swi4 swi6 [pGAL::SWI4] strain (BY147) dies on glucose-containing media due to inadequate expression of G1 cyclins (NASMYTH and DIRICK 1991 ), we found that the same strain transformed with pNM1 was alive on glucose-containing media, indicating that the residual low level of Swi6 produced from pNM1 was sufficient to complement deletion of swi6 in a swi4 mutant background. Given this observation, we mutagenized swi6 [pNM1] cells and screened for Swi6-dependent viability in two ways: (1) on glucose-containing medium to decrease expression of SWI6 from the GAL promoter on pNM1 to very low levels and (2) on 5-FOA-containing medium to select for cells that had lost the URA3-containing plasmid (BOEKE et al. 1984 ) and so completely eliminate all Swi6 coding sequences from the cell.

Initial characterization of slm mutants:
We screened ~40,000 colonies from 12 independent pools of strain NY3 treated with EMS for their ability to grow on YEPD (glucose-containing) or 5-FOA-containing medium. Colonies were also tested for their ability to grow at both high (37°) and low (25°) temperatures to allow for identification of temperature-sensitive slm mutants. In this way, a total of 244 colonies were identified that failed to grow on YEPD or 5-FOA at 25° or 37°. Of these mutant isolates, 58 were defective in glucose metabolism, 20 carried mutations conferring temperature-sensitive lethality that was independent of Swi6, and 4 were unable to grow on medium lacking uracil (see MATERIALS AND METHODS). The remaining 162 isolates showed different minimal conditions required for lethality (Fig 1, data not shown). We mated the 162 strains to the parental swi6 strain and found that all but one of the mutations tested were recessive.

To determine the number of loci that were represented by the 161 recessive SWI6-requiring mutant strains, a complementation test was performed. Diploids created by pairwise crossing of the haploid mutants (see MATERIALS AND METHODS) were tested for the ability to grow in the absence of Swi6. Of the 161 slm strains, 128 fell into 10 complementation groups designated SLM1 through SLM10. The 34 mutants that did not fit into these 10 groups were either found in multiple complementation groups or were not members of any of the 10 complementation groups. Representatives of each complementation group were backcrossed to an isogenic swi6 [pNM1] strain. The resulting strains containing mutations in SLM1 through SLM10 were designated NYS1 through NYS10 (Fig 1).

Rescue of slm mutants by SWI4:
We expected to isolate mutants defective in SWI4 in our screen, since swi4 swi6 strains are inviable (BREEDEN and NASMYTH 1987A ). To assay our slm isolates for potential SWI4 mutations, all slm swi6 strains were tested for suppression of the swi6-dependent synthetic lethality by transformation with a plasmid carrying the SWI4 gene (Fig 2 and data not shown). Viability was restored to both NYS4 and NYS6 when they were transformed with a centromere-derived plasmid containing the SWI4 gene, but not when transformed with vector alone (Fig 2). To test whether the slm4 mutant harbored a mutation in SWI4, we crossed a slm4 swi6 [pNM1] strain to a swi4 SWI6 strain (NYS4 x BY167) and subjected the diploids to tetrad analysis. Of 10 tetrads analyzed, all had two viable and two inviable meiotic segregants when tested on 5-FOA-containing medium to select against the plasmid-borne SWI6 gene (Table 2). This 2:2 segregation pattern, which reflects lack of recombination between swi4 and slm4, showed that SLM4 and SWI4 were indeed allelic.

We performed a similar segregation test with a slm6 mutant strain by crossing a slm6 swi6 [pNM1] strain to the swi4 SWI6 strain (NYS6 x BY167). In this case, 12 tetrads were recovered on galactose-containing medium and meiotic segregants were tested for viability on 5-FOA-containing medium at 25° (the restrictive conditions for the slm6 allele used in the cross). Some of the tetrads yielded progeny with 3:1 and 4:0 segregation of viability:lethality, indicating the recovery of SLM6 and SWI4 recombinants (Table 2). We conclude that, although the slm6 swi6 lethality is rescued by ectopic expression of SWI4, the slm6 strain is mutated at a locus that is unlinked to SWI4.

Cell cycle phenotypes of the slm mutants:
If an alternative pathway for expression of SBF- or MBF-driven genes were impaired in a slm mutant, the swi6 slm double mutant might be expected to show a specific cell cycle-arrest phenotype. Since we were particularly interested in any mutants that might identify pathways that function together with Swi6 to regulate gene expression at START, we used FACS analysis to examine the arrest phenotype of the swi6 slm mutants [slm4 (swi4) and mutants were not examined; slm2 and slm5 mutant strains were extremely clumpy and reproducible FACS profiles were difficult to obtain]. Four slm mutants do not show a cell cycle-specific arrest: slm1, slm3, slm9, and slm10 (data not shown). Neither the FACS profiles nor the budding index of these mutants was altered when cells were arrested under nonpermissive conditions (data not shown).

Three other slm swi6 mutants did show a cell cycle-arrest phenotype when incubated under restrictive conditions. Both slm7 and slm8 strains showed a uniform morphological arrest with 85–92% large, unbudded cells in the cultures after an 8-hr incubation in the nonpermissive conditions (467/508 unbudded cells for slm7 and 435/514 unbudded cells for slm8; Fig 3A, slm7-1 swi6, slm8-1 swi6). Arrest as large, unbudded cells is typical of cells unable to execute START, such as those deleted for both SWI4 and SWI6 (swi4 swi6, Fig 3A).

Although the morphological arrest phenotype of the slm7 and slm8 cells was uniform, the FACS profile of the slm7-1 swi6 and slm8-1 swi6 mutants under nonpermissive conditions showed some cells in G2. The nonuniformity of the arrest may reflect residual SWI6 or SLM gene function allowing DNA replication in some cells in the culture or it may reflect an uncoupling of the bud morphogenesis and DNA replication pathways in cells lacking Swi6 and Slm7 or Slm8. In any case, the arrest phenotype is consistent with a redundant role for SLM7 and SLM8 with SWI6 in regulating G1 progression. DAPI staining did not reveal any obvious nuclear defects in arresting swi6 slm7 or swi6 slm8 cells.

In contrast to slm7 and slm8, the FACS profile of the slm6 swi6 double mutants showed a shift to a larger G2 peak under nonpermissive conditions (Fig 3A, slm6-1 swi6). After 4 hr in arrest conditions, some of the NYS6 cells were multiply budded with only one or two nuclei visible by DAPI staining. After 8 hr, the majority of the swi6 slm6 cells were arrested as uninucleate cells with elongated buds (Fig 3A, slm6-1 swi6, 8 hr). Some of these cells did not appear to have nuclei, even at permissive temperatures, indicating a possible nuclear migration defect in swi6 slm6 cells.

For cell cycle analysis of the slm swi6 [pGAL::SWI6] mutants, there were two changes to the growth conditions: the carbon source for the cells was changed from galactose to glucose and the temperature was raised from 25° to 37°. To eliminate the effects of changing both the carbon source and the level of Swi6 protein, we examined the slm mutants in an otherwise wild-type background (slm SWI6). We recovered meiotic segregants from diploids formed by mating NYS6, 7, and 8 to wild-type (JO34) cells. The slm6-1 SWI6 strain (NY7) was not temperature sensitive and showed no cell cycle-specific defects (Fig 3B). In contrast, spores that carried a wild-type SWI6 gene (His^-) but were still temperature sensitive for viability (on YEPD at 37°) were recovered with a frequency consistent with slm7-1 and slm8-1 strains having a temperature-sensitive growth defect in an otherwise wild-type (SWI6) background. For the slm7-1 strain, this result was not unexpected since the slm7-1 swi6 isolates showed synthetic lethality at low temperature (25°) and were not originally screened for temperature sensitivity on galactose-containing medium (see MATERIALS AND METHODS and Fig 1). However, the slm8 swi6 mutant showed synthetic lethality only at high temperature and the double mutant strain grew well on galactose-containing medium at 37° (see Fig 1). We conclude that the temperature-sensitive phenotype of the slm8-1 mutant strain was suppressed by overexpression of SWI6 on galactose-containing medium at 37° (Fig 1, NYS8). We also saw this suppression when a high-copy (2µ) plasmid containing SWI6 was used to transform the slm8 strain (data not shown).

When incubated at the restrictive temperature, the slm7-1 SWI6 strain (NY11) showed a similar FACS profile to the wild-type control strain (JO34, Fig 3B), and no uniform morphological arrest was observed. Therefore, the G1 arrest phenotype that we observe in the slm7-1 swi6 double mutant at low temperature requires mutation of both slm7 and swi6. By contrast, FACS analysis of a culture of slm8-1 SWI6 cells (NY15) showed an accumulation of cells with 1N DNA content, with the majority of the cells arresting before 2 hr (Fig 3B). This 1N accumulation was not accompanied by a morphological arrest, as many of the cells continued to bud and did not become appreciably larger during continued incubation at the restrictive temperature. We conclude that mutation of SLM8 causes a defect in S-phase progression that results in a G1 arrest phenotype in the absence of Swi6.

Analysis of cell cycle-regulated gene expression in a slm7 mutant:
In our synthetic lethal screen with SWI6, we expected to isolate genes that, like SWI6, were involved in controlling cell cycle-regulated transcription. We therefore examined the transcript levels of cell cycle-regulated genes in slm6-1, slm7-1, and slm8-1 strains. We analyzed gene expression at the permissive temperature (25°) since defects in gene expression seen under permissive growth conditions are likely to reflect a true gene expression defect as opposed to a secondary phenotype due to growth arrest.

Yeast cells were arrested at START with mating pheromone (-factor), released into fresh medium, and RNA was prepared from samples of the cells as the culture progressed synchronously through the cell cycle. To accommodate the longer cell cycle in the slm7 mutant, we harvested cells for a longer period of time and sampled every 15 min, instead of every 10 min for the wild-type culture. Northern blot analysis of RNA samples showed that only the slm7-1 SWI6 strain displayed defects in cell cycle-regulated transcription (Fig 4 and data not shown for slm8-1, slm6-1). The maximal level of expression of the cell cycle-regulated gene PCL2, whose expression depends on SBF, and of CLB5, whose expression depends on MBF, was reproducibly reduced at START in cells containing the slm7-1 mutation (peak expression reduced up to 2.5-fold for CLB5 and 2.3-fold for PCL2). However, the maximal level of expression of the CLN2, PCL1 (SBF-dependent), and PCL9 (Swi5-dependent) genes was normal (Fig 4, data not shown for PCL1). We also examined cell cycle-regulated expression of the G2-regulated gene SWI5 to control for changes in synchrony between the wild-type and slm7 SWI6 mutant cultures. The levels of SWI5 transcript indicated that both the wild-type and slm7 SWI6 cultures progressed synchronously through two cell cycles, although some loss of synchrony was observed in the second cell cycle in the slm7-1 SWI6 culture.

View larger version (78K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Northern analysis of transcripts in NY11 (slm7-1) and JO34 (wild-type) cells following release into the cell cycle after -factor block. (A) Aliquots of wild-type cells were taken every 10 min and aliquots of slm7-1 SWI6 cells were taken every 15 min following release from -factor block. The blots were probed for PCL2 and SWI5 transcripts, stripped and probed for CLN2 and ACT1 transcripts, and then stripped and probed for PCL9 transcripts. A separate set of blots was probed for CLB5 transcripts, and then stripped and probed for ACT1 transcripts. The membranes containing NY11 and JO34 RNA were probed under identical conditions with identical probes. The blots shown are representative of several independent experiments. (B) Quantification of the transcripts in A. All bands were normalized to the ACT1 transcript. The intervals at which samples were taken following release from cell cycle block are indicated on the abscissa. The upper values of the abscissa indicate the times when samples were taken from cells containing the slm7-1 mutation (slm7), whereas the lower values indicate the times when samples were taken from wild-type cells (wt).

A slm7-1 swi6 [pNM1] strain dies in the presence of 5-FOA at 25° (Fig 1) and arrests in G1 phase at high temperature (Fig 3A). However, a slm7-1 SWI6 cell does not show a cell cycle-specific defect at 37° (Fig 3B). It is possible that the lack of a cell-cycle specific arrest by the slm7-1 SWI6 at the restrictive temperature indicates that slm7-1 mutants have a general transcriptional defect, not a specific defect in cyclin transcription. To test whether a slm7-1 strain was generally defective in transcription, wild type (BY109), slm7-1 (NY11), taf145^ts (YSW93), and rbp1-1 (Y260) strains were grown to log phase at 30° and shifted to 37°. RBP1 encodes the largest subunit of RNA polymerase II and cells containing the rbp1-1 mutation show a rapid decrease in overall transcript levels at the restrictive temperature. By contrast, cells carrying a temperature-sensitive allele of the gene encoding the TATA-binding protein (TBP)-associated factor Taf145 (taf145^ts) do not have a general transcriptional defect but do have a specific defect in cyclin transcription (WALKER et al. 1997 ). Total RNA was extracted from cells taken from the arresting cultures at 0, 1, 2, and 4 hr after shift to the nonpermissive temperature. Levels of total polyA⁺ mRNA were assessed by hybridization of ³²P-labeled oligo(dT)₂₀ to 10 µg of total RNA spotted on a nylon membrane (Fig 5). While the rbp1-1 strain showed a rapid decrease in general transcription, slm7-1 and taf145^ts strains did not show an equivalent drop in transcript levels. This result suggests that the temperature sensitivity of the slm7-1 strain is not due to a general defect in gene expression but rather due to specific deregulation of a subset of important genes that includes CLB5 and PCL2.

View larger version (35K): [in this window] [in a new window] [Download PPT slide]   Figure 5. RNA polymerase II-dependent transcription in slm7-1. Wild type, slm7-1, taf145ts, and rbp1-1 strains were grown to log phase at 30° and shifted to 37°. RNA samples were taken at 0, 1, 2, and 4 hr after temperature shift. A total of 10 µg of total RNA was dot blotted onto a nylon membrane and probed with 32P-labeled oligo(dT)20 as in WALKER et al. 1997 .

Rescue of slm7-1:
We were interested in identifying the mutated genes in those slm strains that showed a defect in cell cycle-regulated transcription and we focused our attention on SLM7. To clone SLM7, we transformed the slm7-1 (NY11) strain with a genomic library (ENGEBRECHT et al. 1990 ) and identified 17 plasmids that allowed slm7-1 cells to grow at 37°. Although not all of the NY11-rescuing plasmids were identical, PCR analysis revealed that all of them contained the gene encoding the TBP-associated factor Taf17 (data not shown). We subcloned a fragment encompassing the entire open reading frame of TAF17 and some adjacent genomic DNA into a yeast vector (pNM10). The plasmid carrying the subcloned fragment rescued the temperature-sensitive phenotype of slm7-1 (Fig 6) and the synthetic lethality in a slm7-1 swi6 background (data not shown). Confirmation that the rescue was attributable to the TAF17 gene and not other genomic sequences was provided by expressing TAF17 from the galactose-inducible GAL1-10 promoter (MOQTADERI et al. 1996B ). This pGAL::TAF17 plasmid allowed growth of a slm7-1 strain on galactose at 37° (Fig 6) but not on glucose at 37°, indicating that expression of TAF17 can rescue the temperature-sensitive growth defect in the slm7 strain.

View larger version (56K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Rescue of slm7-1 by TAF17. NY11 (slm7-1 SWI6) transformed with vector, pNM10 (pTAF17) and pZM255 (pGAL::TAF17), grown at 30° or 37° on glucose or galactose. Log phase cultures were diluted to an OD600 of 0.1 and serial dilutions were spotted on each plate. Spots were grown for 2 days on glucose and 6 days on galactose at 30° and 37°.

To ask if slm7-1 corresponds to a mutant allele of TAF17, we used an integrating plasmid to target LEU2 to the TAF17 locus without disrupting the gene. The TAF17::LEU2 strain (NY36) was crossed to a slm7-1 strain (NY34). The TAF17::LEU2/slm7-1 diploid was sporulated and the progeny of the resulting tetrads tested for growth on media lacking leucine (caused by the LEU2 gene) and for death on SD at 37° (caused by the slm7-1 mutation). Of 26 tetrads tested, no progeny showed cosegregation of LEU2 and slm7, indicating that SLM7 is allelic to TAF17.

To identify the TAF17 mutation in the slm7-1 strain, we used whole cell PCR to amplify the TAF17 gene from strain NY11 (slm7-1). Both strands of three different PCR fragments were sequenced and the resulting sequences compared to the Saccharomyces Genome Database sequence. All six sequences were in agreement and showed that the slm7-1 mutation changed codon 133 of TAF17 from a tryptophan codon (TGG) to an amber stop codon (TAG). The mutated gene is predicted to encode a Taf17 derivative lacking the C-terminal 25 amino acids, which include a conserved region of 11 amino acids (Fig 7). We conclude that mutation of the Taf17 protein can lead to defects in G1-regulated gene expression and causes cell cycle arrest in G1 phase when Swi6 levels are low and the cells are grown under nonpermissive conditions.

View larger version (70K): [in this window] [in a new window] [Download PPT slide]   Figure 7. Alignment of yeast Taf17slm7-1 with wild-type yeast Taf17 and its human and Drosophila homologues. This alignment was performed with MSA version 2.1 (GUPTA et al. 1995 ). Identical residues are darkly shaded; similar residues are lightly shaded. The histone fold as predicted by XIE et al. 1996 is underlined. The boxed region is the region of greatest similarity among these proteins and is predicted to be missing from the protein encoded by the slm7-1 allele of TAF17. The translational termination signal for Taf17slm7-1 is indicated by the asterisk.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Swi6 is a regulatory component of two transcription factors, SBF and MBF, that promote G1-specific transcription of a number of genes that are essential for START and S-phase in budding yeast (reviewed in BREEDEN 1995 ; KOCH and NASMYTH 1994 ). We have identified several slm mutants (swi6-dependent lethal mutants) that cause inviability in the absence of Swi6. A subset of these slm mutants showed a cell cycle-specific arrest phenotype under restrictive conditions. One of the slm strains contained a mutant allele of SWI4; mutation of SWI4 was known to be synthetically lethal with mutation of SWI6 (BREEDEN and NASMYTH 1987A ) due to inadequate expression of G1 cyclins. The slm7-1 mutant strain harbored a temperature-sensitive allele of TAF17 and arrested in G1 phase of the cell cycle at the nonpermissive temperature when Swi6 levels were low. At the permissive temperature, taf17^slm7-1 SWI6 cells had specific defects in the cell cycle-regulated transcription of a subset of Swi6-dependent target genes. Our data implicate transcriptional regulatory complexes containing Taf17 in cell cycle regulation.

Cell cycle-arrest phenotypes of slm6-1 swi6 and slm8-1 swi6 mutants:
In this screen we hoped to identify genes involved in activating SBF- and MBF-driven gene expression in the absence of SWI6. We identified mutations in three complementation groups, slm6, slm7, and slm8, that died in the absence of Swi6 and had a uniform morphological arrest; the arrest phenotype of these strains was not rescued by overexpression of the cyclins PCL2, CLB5, or CLN2 (data not shown). To date, we have identified the mutated gene in the slm7 strains and we discuss this result below. However, the arrest phenotypes of the slm6 and slm8 strains suggest that further analysis of the SLM6 and SLM8 genes will be of some interest. A slm6-1 mutation caused swi6 cells to accumulate in G2 phase with elongated buds and this arrest could be overcome by increasing the copy number of SWI4. The arrest phenotype of slm6-1 swi6 mutants may reflect a problem with budding or cell wall biosynthesis. Both the SBF and the Pkc1/Slt2 mitogen-activated protein (MAP) kinase pathway are required for maximal induction of cell wall biosynthetic genes encoding proteins used in budding (IGUAL et al. 1996 ). Swi6 is phosphorylated in yeast cells in a Slt2-dependent manner and, like slt2 mutants, the G2 phase arrest of slm6-1 swi6 mutant cells can be overcome by overexpression of SWI4 but not CLN2 (IGUAL et al. 1996 ; MADDEN et al. 1997 ). Perhaps overexpression of SWI4 in the slm6-1 swi6 mutant allows activation of SBF target genes required for cell wall biosynthesis and budding. Allelism of SLM6 with genes encoding components of the Pkc1 pathway has not been tested and mutation of SWI4 or SWI6 is known to cause synthetic lethality with mutation in genes in the PKC-MAP kinase pathway (GRAY et al. 1997 ). However, unlike known alleles of PKC1 and SLT2, the slm6-1 allele does not cause cells to be temperature sensitive in a SWI6 background.

In contrast to slm6, the slm8 mutation caused a temperature-dependent accumulation of cells in G1 and cell death in the absence of SWI6. In addition, mutation of SLM8 in the presence of wild-type SWI6 caused a marked accumulation of cells with 1N DNA content when these cells were shifted from 25° to 37°. The arrest of NY15 cells with G1 DNA content was not accompanied by a morphological G1 arrest. Rather, these cells continued to produce buds, indicating that in slm8 swi6 cells budding and DNA synthesis may be unlinked. At the restrictive temperature, swi4 cells arrest with the opposite phenotype (BY108, Fig 3B), as large unbudded cells with mixed DNA content (OGAS et al. 1991 ; GRAY et al. 1997 ). These phenotypes suggest that Swi4 and Slm8 may be acting in different processes at Start, with Swi4 having a role in budding and Slm8 having a role in DNA replication. Elucidation of the precise relationship between SWI6, SLM6, and SLM8 awaits the cloning and characterization of the SLM6 and SLM8 genes.

slm7-1 is allelic with the gene encoding the histone H3-like protein Taf17:
Complementation and allelism tests revealed that slm7-1 was allelic to TAF17. TAF17 is an essential gene encoding the TBP-associated factor Taf17/Taf20. Taf17 is one of several TBP-associated factors (TAF_IIs) that are subunits of the general transcription factor IID (TFIID). TFIID and other general transcription factors are components of the RNA polymerase II holoenzyme (reviewed in BERK 1999 ). TAF_IIs are highly conserved proteins and all 12 yeast TAF_IIs (yTAF_IIs) have homologues in other eukaryotes, but some metazoans have additional cell type-specific TAF_IIs (reviewed in TANSEY and HERR 1997 ). Roles for TAF_IIs in cell cycle regulation have been suggested by a number of studies in both yeast and mammalian cells. In yeast, transcription of G1 cyclins is dependent on the TAF145 gene; G1 cyclin transcript levels are reduced at the restrictive temperature in a taf145^ts strain, leading to cell cycle arrest in G1 phase (WALKER et al. 1996 , WALKER et al. 1997 ). A role for other TAF_IIs in regulating cell cycle-dependent gene expression is suggested by the G2 phase arrest seen at the restrictive temperature in yeast cells carrying temperature-sensitive alleles of TAF90 or TSM1, a gene that encodes the yeast homologue of the higher eukaryotic TAF_II150 (APONE et al. 1996 ; WALKER et al. 1996 ). TAF_II homologues from different organisms may also be functionally conserved; a temperature-sensitive allele (CCG1) of hTAF250, the mammalian homologue of yTAF145, causes mutant cells to arrest in G1 (SEKIGUCHI et al. 1991 ; HISATAKE et al. 1993 ; RUPPERT et al. 1993 ). Also, the human homologue of yeast TSM1, hTAF_II150, appears to be required for the G2/M transition in mammalian cell lines and for activation of B-type cyclin gene expression (MARTIN et al. 1999 ).

yTaf145 has histone acetylase activity and this activity resides in the region most conserved between mammals and yeast (MIZZEN et al. 1996 ). A group of yTaf_IIs—Taf17, 25, 60, 61, and 90—copurifies with another histone acetylase complex called SAGA (GRANT et al. 1997 , GRANT et al. 1998 ). A subset of these TAF_IIs (yTaf17, Taf60, and Taf61) have homology to histone proteins (H3, H4, and H2B, respectively; reviewed in BURLEY and ROEDER 1996 ) and may form a histone-like octamer (HOFFMANN et al. 1996 ; NAKATANI et al. 1996 ; XIE et al. 1996 ). Consistent with the discovery of common TAF_IIs in TFIID and SAGA, TAF_IIs may promote transcriptional activation by transcription factors either through TFIID or the SAGA complex (KLEMM et al. 1995 ; DRYSDALE et al. 1998 ). The isolation of mutations in TAF17 in our screen may be a reflection of its role in either TFIID or the SAGA histone acetyl transferase complex.

slm7-1 (taf17) mutants show transcriptional defects in cell cycle-regulated genes:
Cells containing temperature-sensitive mutations in some Taf_IIs show broad defects in transcriptional activation under nonpermissive conditions, but these defects are not global (APONE et al. 1996 , APONE et al. 1998 ; SHEN and GREEN 1997 ; WALKER et al. 1997 ; HOLSTEGE et al. 1998 ; MOQTADERI et al. 1998 ; NATARAJAN et al. 1998 ). Rather, mutant Taf_IIs cause gene-specific and Taf-specific defects in transcription under nonpermissive conditions (WALKER et al. 1997 ; WANG et al. 1997 ; HOLSTEGE et al. 1998 ). For example, as noted above, taf145^ts mutants are defective in transcription of a specific subset of genes, the G1 cyclin-encoding genes (WALKER et al. 1997 ), and genome-wide analysis of transcript levels in taf145^ts mutants reveals that 16% of genes examined are as dependent on Taf145 as they are on the largest subunit of RNA polymerase II (HOLSTEGE et al. 1998 ). Our discovery that taf17^slm7-1 mutants have specific gene expression defects at the permissive temperature, and have a cell cycle-arrest phenotype at the nonpermissive temperature when Swi6 levels are low, provides another example of a Taf that may play a role in cell cycle-dependent gene expression.

Genome-wide assessment of transcript levels using DNA microarrays in strains carrying temperature-sensitive mutations in Taf17 shows that Taf17 is required for transcription of ~67% of genes tested under restrictive conditions (APONE et al. 1998 ; HOLSTEGE et al. 1998 ). A third study using Taf17-depleted cells concluded that most promoters were Taf17-dependent (MOQTADERI et al. 1998 ). Whereas these studies examined transcript levels in heat-treated, dying cells, our study examined cell cycle-regulated genes in a taf17^ts mutant at the permissive temperature; under these conditions, the cells remained viable and were not heat stressed. Also, the TAF17 alleles used in other studies contained mutations in the histone fold (APONE et al. 1998 ; MICHEL et al. 1998 ), which is necessary for interaction with other histone-like Taf_IIs (HOFFMANN et al. 1996 ; NAKATANI et al. 1996 ; XIE et al. 1996 ; APONE et al. 1998 ; MICHEL et al. 1998 ). Many of the TAF_IIs examined under nonpermissive conditions in taf145^ts or taf17^ts cells were unstable (WALKER et al. 1997 ; APONE et al. 1998 ; HOLSTEGE et al. 1998 ). This may be because the temperature-sensitive mutations in Taf145 or Taf17 prevented formation of a Taf-containing complex at the nonpermissive temperature, causing the constituent proteins of this complex to be accessible to the degradation machinery.

The stability of the Taf17^slm7-1 protein has not been established, so the defects in transcription observed in taf17^slm7-1 cells may be due to degradation of Taf17^slm7-1 and associated proteins and not due to the specific mutation in Taf17^slm7-1. However, the taf17^slm7-1 mutation is predicted to encode a Taf17 derivative with an intact histone fold, but with a deletion encompassing a highly conserved region outside the histone fold. Since the histone fold of Taf17^slm7-1 is intact, this Taf17 variant may interact normally with the other components of the histone-like Taf octamer, and its association with other proteins may promote the stability of Taf17-containing complexes.

The specific defects in cell cycle-regulated transcription in taf17^slm7-1 cells suggest that the Taf17^slm7-1 protein may only be of regulatory importance in the context of certain promoters. Our data suggest that the CLB5 and PCL2 promoters are partially Taf17 dependent and that CLN2, PCL1, and PCL9 expression is relatively independent of Taf17. Our results agree in part with the DNA microarray analysis on the taf17^ts mutant strains described above (http://gaiberg.wi.mit.edu/cgi-bin/young_public/factor.cgi?gene=TAF17) (HOLSTEGE et al. 1998 ). In these experiments, CLB5 transcription decreased 8.7-fold in the taf17^ts cells after 45 min at the restrictive temperature, whereas CLN2 and PCL9 transcription decreased only 1.1- and 1.5-fold, respectively. Reproducible results for PCL2 were not obtained. We did not observe induction of PCL1 in taf17^slm7-1 cells at 25° but in the gene-chip analysis, PCL1 was induced 3.2-fold under nonpermissive conditions. PCL1 is induced at high temperature (MADDEN et al. 1997 ), so the induction observed by Holstege et al. may occur only at 37°. The DNA microarray studies do not clarify whether the pattern of gene expression we see in the taf17^slm7-1 mutant is due to its role in TFIID or its role in the SAGA complex.

Reasons for synthetic lethality of mutations in TAF17 and mutations in SWI6:
We consider two possible explanations for the synthetic lethality of mutations in SWI6 and mutations in TAF17. Either SCB and/or MCB-driven transcription is Taf17 dependent, or SCBs and MCBs are Taf17 independent and Taf17 controls cell cycle-regulated transcription through other elements. If SCBs and MCBs are Taf17 dependent, the synthetic lethality of mutations in TAF17 and mutations in SWI6 may be due to reduced induction of SBF- or MBF-dependent genes. In swi6 cells, induction of SCB-containing promoters at Start is reduced relative to wild type (NASMYTH and DIRICK 1991 ; OGAS et al. 1991 ; DIRICK et al. 1992 ; LOWNDES et al. 1992 ; BREEDEN and MIKESELL 1994 ; CROSS et al. 1994 ; STUART and WITTENBERG 1994 ). The residual levels of induction of essential SBF-dependent genes in swi6 cells may require Swi4, since elimination of Swi4 from the cell causes cell death (BREEDEN and NASMYTH 1987A ). In swi6 cells, induction of MCB-containing genes is decreased, but the basal transcription level of these genes is intermediate between activated and inactivated levels (KOCH et al. 1993 ). Mutation of Taf17 may result in a reduction of this basal level, which could compromise cell viability.

If SCBs and MCBs are Taf17 independent, there are two possibilities that may explain the synthetic lethality of mutations in SWI6 and mutations in TAF17. First, SBF- or MBF-dependent genes may contain Taf17-dependent elements that are distinct from SCBs and MCBs. Alternatively, Taf17 may activate transcription of SBF- or MBF-independent genes that are necessary for progression through Start in the absence of Swi6. In both these cases, the synthetic lethality of mutations in TAF17 and mutations in SWI6 may be due to a defect in cell cycle regulation by other transcriptional activators. An example of an SBF- and MBF-independent element that could be Taf17 dependent is the ECB promoter element. ECB elements are found upstream of cell cycle-regulated genes, including SWI4, and are under the control of the Mcm1 protein and perhaps other proteins (MCINERNY et al. 1997 ). Effects on SWI4 expression are unlikely to account entirely for the phenotype of a taf17^slm7-1 strain, since expression of SWI4 from a heterologous promoter failed to rescue the synthetic lethality of a taf17^slm7-1 swi6 strain (data not shown and RESULTS).

Role of chromatin-remodeling complexes in transcriptional activation by SBF or MBF:
Recent chromatin immunoprecipitation experiments have shown that two chromatin remodelling complexes, SAGA and Swi/Snf, facilitate the binding of SBF to the HO promoter in G1 phase (COSMA et al. 1999 ). Our discovery of a genetic interaction between Taf17 and Swi6 may reflect a requirement for chromatin remodelling factors in the proper cell cycle regulation of other SBF- and MBF-dependent genes. There are several ways that the Taf17-containing SAGA or TFIID complexes may assist in the activation of RNA polymerase by SBF or MBF. The presence of the coactivator or the disruption of histone-DNA contacts may encourage binding of SBF or MBF to target sites. Alternatively, coactivators might assist transcriptional activation through binding of SBF or MBF to TFIID or SAGA; this binding may occur through a direct interaction with the Taf17/Taf60/Taf68 histone-like octamer. Since combining the defects of taf17 mutants and swi6 mutants results in death, they are not epistatic to each other. Binding of SBF and MBF to Taf17 in a wild-type cell may therefore occur through Swi4 or Mbp1. Support for a direct binding model comes from studies of other transcription factors. The Gcn4 transcriptional activator binds components of both TFIID and SAGA, including Taf17, and the human and Drosophila homologues of Taf17 can bind to the VP16 and p53 transcriptional activators (GOODRICH et al. 1993 ; LU and LEVINE 1995 ; THUT et al. 1995 ; DRYSDALE et al. 1998 ). Also, the histone H2B-like protein Taf61 is required by Gcn4 to recruit the SAGA complex (NATARAJAN et al. 1998 ).

In summary, we have found mutations in 10 genes, SLM1–10, that are synthetically lethal with deletion of SWI6. Mutations in one of the SLM genes, SLM6, caused swi6 cells to accumulate in G1 phase and may lead to a defect in budding or cell wall biosynthesis. Two mutations, in slm7 and slm8, caused accumulation in G1 phase in a swi6 mutant background. The slm7-1 m